Chapter 4: Preprocessing¶
Continuing from Chapter 3 — Importing Data.
Raw recordings are noisy before they're useful — the same preprocessing goals as the MNE-Python track's Preprocessing chapter, done here by clicking rather than by writing raw.filter(...).
Marking bad channels and segments¶
Open the recording in the Viewer and you can click directly on a channel's label to mark it "bad" (excluded from later analysis), or drag-select a noisy stretch of time to mark as a bad segment. Both are stored as metadata on the file — nothing is deleted, so you can always change your mind.
Running a process¶
This is the pattern you'll repeat constantly in Brainstorm, so it's worth naming explicitly:
- Select the file (or files) in the Database Explorer tree.
- Drag them into the Process1 tab.
- Type into the process search box — e.g. "band-pass" — and pick the matching process from the list.
- Set its parameters in the dialog that appears (for a filter: low cutoff, high cutoff, exactly like MNE-Python's
l_freq/h_freq). - Click Run.
The result appears as a new file in the tree, alongside the original — filtering never overwrites your import, the same "leave the original untouched" principle as raw.copy().filter(...) in MNE-Python.
Common preprocessing processes¶
- Band-pass / notch filters — under the "Pre-process" category, remove slow drift, high-frequency noise, and power-line interference.
- SSP and ICA — under "Artifacts", the same statistical techniques MNE-Python's ICA chapter uses to separate out eye-blink and heartbeat artifacts from brain signal, run here through a guided dialog instead of code.
- Re-referencing — changing which channel(s) EEG voltages are measured relative to, the same concept covered conceptually in the MNE-Python track's EEG Fundamentals chapter.